Publications

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

Recent years have witnessed great success of deep convolutional networks in sensor-based human activity recognition (HAR), yet their practical deployment remains a challenge due to the varying computational budgets required to obtain a reliable prediction. This paper focuses on adaptive inference from a novel perspective of signal frequency, which is motivated by an intuition that low-frequency features are enough for recognizing “easy” activity samples, while only “hard” activity samples need temporally detailed information. We propose an adaptive resolution network by combining a simple subsampling strategy with conditional early-exit. Specifically, it is comprised of multiple subnetworks with different resolutions, where “easy” activity samples are first classified by lightweight subnetwork using the lowest sampling rate, while the subsequent subnetworks in higher resolution would be sequentially applied once the former one fails to reach a confidence threshold. Such dynamical decision process could adaptively select a proper sampling rate for each activity sample conditioned on an input if the budget varies, which will be terminated until enough confidence is obtained, hence avoiding excessive computations. Comprehensive experiments on four diverse HAR benchmark datasets demonstrate the effectiveness of our method in terms of accuracy-cost tradeoff. We benchmark the average latency on a real hardware.

This paper presents the design and implementation of a new modular self-reconfigurable robot, called Transmote. The proposed robot has several novel features that make it suitable for search and rescue operations in complex environments. A single module can move independently by coordinating three joints. Multiple modules can connect with each other to form versatile assembled robotic structures with more powerful locomotion capabilities. The assembled structures can also transform between each other or fully disassemble into individual modules again. The modules, which communicate with each other through ZigBee compliant protocols, can build emergency communication and monitoring networks when spread to the areas without communication infrastructures. Preliminary experimental results demonstrate locomotion, mechanical docking, and transformation of several example configurations.

The tendon–sheath system (TSS) has emerged as a promising solution for force transmission in robot applications, owing to the lightweight, low inertia, and good flexibility. However, the performance of tension control is affected by the configuration-dependent nonlinearities, and achieving precise control of the distal force of TSS in time-varying configurations remains a challenge. To address this issue, this article proposes a comprehensive method combining feedforward and feedback control to enhance the control accuracy of TSS by estimating the bending angle in real time, even when the system configuration is unknown. A bending angle estimation approach based on a friction model is proposed, and an improved friction model is developed to simulate the nonlinear friction more accurately. Furthermore, a double-TSS transmission structure is designed to facilitate the integration of control and configuration feedback in mechanical systems. The experimental results confirm the effectiveness of the proposed method in enhancing force-tracking performance in unknown configurations.

It is an emerging frontier of research on the use of neural signals for prosthesis control, in order to restore lost function to amputees and patients after spinal cord injury. Compared to the invasive neural signal based brain-machine interface (BMI), a non-invasive alternative, i.e., the electroencephalogram (EEG)-based BMI would be more widely accepted by the patients above. Ideally, a real-time continuous neuroprosthestic control is required for practical applications. However, conventional EEG-based BMIs mainly deal with the discrete brain activity classification. Until recently, the literature has reported several attempts for achieving the real-time continuous control by reconstructing the continuous movement parameters (e.g., speed, position, etc.) from the EEG recordings, and the low-frequency band EEG is consistently reported to encode the continuous motor control information. Previous studies with executed movement tasks have extensively relied on the amplitude representation of such slow oscillations of EEG signals for building models to decode kinematic parameters. Inspired by the recent successes of instantaneous phase of low-frequency invasive brain signals in the motor control and sensory processing domains, this study examines the extension of such a slow-oscillation phase representation to the reconstructing two-dimensional hand movements, with the non-invasive EEG signals for the first time. The data for analysis are collected on five healthy subjects performing 2D hand center-out reaching along four directions in two sessions. On representative channels over the cortices encoding the execution information of reaching movements, we show that the low-delta EEG phase representation is characterized by higher signal-to-noise ratio and stronger modulation by the movement tasks, compared to the low-delta EEG amplitude representation. Furthermore, we have tested the low-delta EEG phase representation with two commonly used linear decoding models. The results demonstrate that the low-delta EEG phase based decoders lead to superior performance for 2D executed movement reconstruction to its amplitude based counterparts, as well as the other-frequency band amplitude and power based features. Thus, our study contributes to improve the movement reconstruction from EEG by introducing a new feature set based on the low-delta EEG phase patterns, and demonstrates its potential for continuous fine motion control of neuroprostheses.