Search

Motion intention detection is fundamental in the implementation of human-machine interfaces applied to assistive robots. In this paper, multiple machine learning techniques have been explored for creating upper limb motion prediction models, which generally depend on three factors: the signals collected from the user (such as kinematic or physiological), the extracted features and the selected algorithm. We explore the use of different features extracted from various signals when used to train multiple algorithms for the prediction of elbow flexion angle trajectories. The accuracy of the prediction was evaluated based on the mean velocity and peak amplitude of the trajectory, which are sufficient to fully define it. Results show that prediction accuracy when using solely physiological signals is low, however, when kinematic signals are included, it is largely improved. This suggests kinematic signals provide a reliable source of information for predicting elbow trajectories. Different models were trained using 10 algorithms. Regularization algorithms performed well in all conditions, whereas neural networks performed better when the most important features are selected. The extensive analysis provided in this study can be consulted to aid in the development of accurate upper limb motion intention detection models. Agency for Science, Technology and Research (A*STAR) Published version This work was partially supported by the grant “Intelligent Human-Robot interface for upper limb wearable robots” (Award Number SERC1922500046, A*STAR, Singapore).

The material removal in complicated geometries is the principal objective for machining with compliant abrasive tools in aerospace industries. Realizing ideal material removal rates with fine tolerance in tertiary finishing process such as abrasive belt grinding is essential. This makes it fundamental to look in more detail at the process parameters/variables that affect the material removal rate. However, the relationship between the material removal rate and process parameters is not well understood. Previously, five parameters such as belt speed, feed, rubber hardness, grit size and force applied were studied in correspondence with the depth of cut, and it was found grit size plays a dominant role in the grinding process [1]. In this study, the influence of four parameters out the five parameters namely belt speed, feed, rubber hardness and force are investigated using a dynamic pressure sensor. Three level of input for each parameter was considered. Experimental trials were conducted by varying the levels of one parameter and maintaining a constant level for other three parameters. Based on the experimental trials performed using the dynamic pressure sensor, a correlation between the three levels considered for each parameter is identified based on the contact conditions. It was observed that pressure distribution based on the contact condition using the pressure sensor for the parameter considered followed the same results as predicted by ANOVA [1]. This research work describes a systematic approach to analyse process parameters based on contact conditions using a pressure sensor to understand material removal in a compliant abrasive belt grinding process. NRF (Natl Research Foundation, S’pore) Accepted version

One of the key components in real-time monitoring and control on machining and surface finishing processes are sensors. The advances of such system have triggered interesting questions on sensor selection that act as the fundamental before starting a project. This paper is made to review and answer the questions surrounding sensor selection. The paper first explains on the type of sensors commonly used in practice for real-time monitoring and control systems. After which, the paper discusses on how often the sensors are used on several machining and surface finishing processes and what are the reasons for the sensor selection. Thereafter, a review on the type features commonly analysed through these sensors is discussed. The paper expects reader would decide better upon selecting sensors and has a better direction in their project. Thus the paper works to guide reader to improve based on what has been completed before. Published version

The monitoring of surface finishing processes in aerospace manufacturing industry becomes one of key issues to maintain the overall quality of product or part. To date, the surface quality monitoring post machining processes such as deburring, use visual inspection, surface roughness test or laser gap gun. The whole manufacturing process then requires a considerable amount of time as the production line must be halted due to these measurements taking place. This study presents an online monitoring system to measure the chamfer quality of replica boss hole post-deburring process. Vibration signal was measured during the deburring process and the features that correlate to the deburring stages (passes) were extracted. This paper focuses on the validation of actual vibration signal with the modal analysis of work coupon (replica boss hole) to obtain the correlation between the vibration amplitude level on particular region and the mode shape of work coupon during the deburring process. Published version

Process monitoring using indirect methods relies on the usage of sensors. Using sensors to acquire vital process related information also presents itself with the problem of big data management and analysis. Due to uncertainty in the frequency of events occurring, a higher sampling rate is often used in real-time monitoring applications to increase the chances of capturing and understanding all possible events related to the process. Advanced signal processing methods are used to further decipher meaningful information from the acquired data. In this research work, power spectrum density (PSD) of sensor data acquired at sampling rates between 40-51.2 kHz was calculated and the corelation between PSD and completed number of cycles/passes is presented. Here, the progress in number of cycles/passes is the event this research work intends to classify and the algorithm used to compute PSD is Welch's estimate method. A comparison between Welch's estimate method and statistical methods is also discussed. A clear co-relation was observed using Welch's estimate to classify the number of cycles/passes. The paper also succeeds in classifying vibration signal generated by the spindle from the vibration signal acquired during finishing process. ispartof: SENSORS vol:17 issue:6 ispartof: location:Switzerland status: published